Our Work
Systems that passed compliance review and stayed in production, at institutions that name themselves.
Flagship — Legal
A&GEL is a suite of knowledge augmentation tools, co-developed with A&G Asia and built around how their lawyers already work.
A&G Asia is a network of more than 800 lawyers across six countries and over 40 practice areas. They are the anchor user of A&GEL across six Asia-based offices, which means the product is shaped by daily use rather than by a requirements document.
The models were available to every firm in the market. What could not be licensed was how these lawyers construct an argument.
Verified
Award
Innovation in Digital Legal Practice, at the Financial Times Innovative Lawyers Asia-Pacific Awards 2026, for a fully on-premise generative AI solution.
Allen & Gledhill ↗Build
Two years, co-developed with A&G Asia under the firm's own generative AI taskforce. Their legal methodology and prompt library, our engineering and platform.
Press
Covered by The Business Times in its reporting on generative AI at Allen & Gledhill with IMDA.
The Business Times ↗Trusted and tiered folders
Answers are drawn from private documents, matter folders, firm-wide precedents and selected third-party material, under permissioned access rather than a single undifferentiated pile.
Active work surfaces
Review, Comparison and Workstudio handle document work at scale, so the system supports how a matter is actually run rather than offering a chat box and hoping.
Verification surfaces
Source references, highlighted citations, side-by-side views and redline outputs give a lawyer several routes back to the underlying document before anything is relied on.
No training on your data. No external data transfer. No exceptions.
A&GEL runs on premise, in a client-controlled private cloud, or as A&GEL Cloud, managed by PAND.AI on Google Cloud in the Singapore region. Every deployment mode runs under a zero-trust architecture with tenant and network isolation, AES-256 encryption at rest and TLS 1.3 in transit, single sign-on, and audit logs the firm can inspect.
Selected Engagements
Allianz · Insurance · Singapore
Allison, the AI assistant we built for Allianz, takes a customer through an insurance claim as a conversation rather than a form. It then integrates into the downstream claims process, including robotic process automation and coordination with authorised workshops.
Bangkok Bank · Banking · Thailand
A digital sales assistant for the bank's sales and relationship managers. It handles Thai, English and sentences that mix the two, the way people actually type. It started in transaction banking and has since expanded into wealth management and bancassurance.
CIMB · Banking · Singapore & Malaysia
EVA answers SME and commercial banking questions and helps a business work out what it qualifies for before it applies.
Our Own
Everything above was built for someone else. GINA was built for us.
PAND.AI operates GINA, a consumer insurance broker on WhatsApp, under our own MAS broking licence. A customer compares quotes, pays and receives a policy without leaving the conversation. It runs on the same platform as everything else on this page.
This is the live product, not a demo.
On the public register
Licensed for general insurance broking by the Monetary Authority of Singapore.
Money moves and policies are issued, so somebody has to answer for both — and that somebody is us.
The engineers behind these systems are the ones you would be working with. Tell us your constraints and we will tell you honestly whether we can help.
Based in Singapore · Serving Southeast Asia